Batch API

Submit large job offline at 50% discount. Completes within 24 hours.

Batch APIs let you submit thousands of requests as a single file, processed asynchronously within a provider-specified SLA (usually 24 hours), at 50% off both input and output prices. Anthropic, OpenAI, and Google all offer this. Use batch for anything that isn't user-facing: evals, offline classification, content generation for blogs, data enrichment. The catch is latency — if your app needs a response in seconds, batch is useless. But for background workloads, batch is the easiest 50% cost cut available.